Re: correcting Emacs CHAR_COMPONENTS_VALID_P

[Dave Love <d.love@dl.ac.uk>]

You wrote: 

> Dave Love <d.love@dl.ac.uk> writes:
> 
> > I accidentally made a bogus character from a 94x94 charset like
> >   (make-char 'japanese-jisx0208 32 32)
> 
> > and was confused by the result of inserting it.
> 
> > Is it worth fixing, or is it intentional for efficiency or something?
> > It isn't trivial to fix because several functions and macros are
> > affected.
> 
> The current behaviour is intentional.  Efficiency is one
> reason.  Another reason is historical one.  There exists
> many files that contains invalid byte sequences such as 0xA0
> 0xA0 (especially EUC-GB).  Before we introduce
> eight-bit-control and eight-bit-graphic, we couldn't decode
> them properly if we treat them as invalid byte sequence.
> Thus I made the rule loosen.  Now, as we have
> eight-bit-graphic, we can make the rule firm, but I think
> it's not worth spending time on it.

I think that deserves commentary in the sources to avoid anyone else
wasting time looking at it.